In this thesis, we propose a novel direct pointing method utilizing the on-screen reflection of commodity screens. A myriad of input methods, using user’s hands or eye gaze, have been proposed to interact directly with a screen. Such direct manipulation methods have several advantages in terms of designing the transfer function and using it with little adaptation time. However, they have side effects due to the two innate properties of a screen : public visibility of a screen; and decreasing performance with the distance between a user and a screen. This study presents a new interaction method using on-screen reflection that overcomes these limitations. When we compared the reflector-based authentication system to the existing touch-based system, we found that the password hacking success rate of the reflector is 20 times lower than the one of touch-based input. Moreover, We found our method with user’s reflection has the significantly lower error rate in far distance compared to widely used gaze input method. We expect our model can be applied to several forms of the computing environment with screens and would provide a better user experience.